Investigating the Behavior of Steel Flush Endplate Connections at Elevated Temperatures Using FEM and ANN

This paper investigates the moment-rotation (M-θ) behavior of flush endplate (FEP) connections at elevated temperatures using the finite element (FE) method and an artificial neural network (ANN). Firstly, a three-dimensional nonlinear FE model of flush endplate connection is carried out and verified with the tests conducted by others using ABAQUS. Then, an extensive database is created by varying several parameters (i.e., the endplate thickness, the bolt row distance, the pitch of bolts, the gage distance, the outer edge distance, the number of bolt rows, the bolt diameter, and the material properties) to get insight into the influences of each parameter on the connection behaviors at elevated temperatures. Additionally, a simple and accurate model with two shape parameters for the M-θ relationship of semi-rigid flush endplate connections at elevated temperatures is proposed based on this database. Accordingly, two shape parameters and the ultimate moment (Mu) of the model are determined using the ANN model. Finally, the performance of the proposed model is verified and has a good agreement with various test data.